1 |
dpavlin |
228 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
# this utility will convert some (or all, defined by regex) |
4 |
|
|
# fields into marc file from one or more CDS/ISIS files |
5 |
|
|
# |
6 |
|
|
# 2004-02-23 Dobrica Pavlinusic <dpavlin@rot13.org> |
7 |
|
|
# |
8 |
|
|
# if ISIS databases are named same as directories in which they |
9 |
|
|
# reside, you can specify just directories (so that shell globing work) |
10 |
|
|
# like this: |
11 |
|
|
# |
12 |
|
|
# ./isis2marc.pl all.marc /mnt2/*/LIBRI |
13 |
|
|
|
14 |
|
|
use strict; |
15 |
|
|
use OpenIsis; |
16 |
|
|
use MARC; |
17 |
|
|
use Data::Dumper; |
18 |
|
|
|
19 |
|
|
# to select all fields use something like /./ |
20 |
|
|
my $field_filter = '^700$'; |
21 |
|
|
|
22 |
|
|
my $marc_file = shift @ARGV || die "Usage: $0 [MARC file] [ISIS db]..."; |
23 |
|
|
|
24 |
|
|
my $marc=MARC->new; |
25 |
|
|
|
26 |
|
|
# it seems that I can't specify invalid template for 005 and prevent |
27 |
|
|
# output from creating field 005 |
28 |
|
|
#$num->add_005s({record=>1}); |
29 |
|
|
|
30 |
|
|
select(STDOUT); $|=1; |
31 |
|
|
|
32 |
|
|
my %stored; |
33 |
|
|
my $total = 0; |
34 |
|
|
|
35 |
|
|
foreach my $db_file (@ARGV) { |
36 |
|
|
|
37 |
|
|
print "reading '$db_file'"; |
38 |
|
|
|
39 |
|
|
if (-d $db_file) { |
40 |
|
|
$db_file =~ s,([^/]+)/*$,$1/$1,; |
41 |
|
|
} |
42 |
|
|
|
43 |
|
|
my $db = OpenIsis::open( $db_file ); |
44 |
|
|
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
45 |
|
|
|
46 |
|
|
print " [rows: $maxmfn]\n"; |
47 |
|
|
|
48 |
|
|
my $progress_len = 50; |
49 |
|
|
|
50 |
|
|
my $step = int($maxmfn/$progress_len); |
51 |
|
|
$step = 1 if ($step == 0); |
52 |
|
|
|
53 |
|
|
my $new = 0; |
54 |
|
|
|
55 |
|
|
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
56 |
|
|
print "." if ($mfn % $step == 0); |
57 |
|
|
my $row = OpenIsis::read( $db, $mfn ); |
58 |
|
|
foreach my $fld (keys %{$row}) { |
59 |
|
|
next if ($fld !~ m/$field_filter/); |
60 |
|
|
|
61 |
|
|
my @values; |
62 |
|
|
my $num; |
63 |
|
|
|
64 |
|
|
foreach my $sf (@{$row->{$fld}}) { |
65 |
|
|
|
66 |
|
|
$stored{$sf}++; |
67 |
|
|
|
68 |
|
|
next if ($stored{$sf} > 1); |
69 |
|
|
|
70 |
|
|
my %v; |
71 |
|
|
while ($sf =~ s/\^(\w)([^\^]+)//) { |
72 |
|
|
$v{$1} = $2; |
73 |
|
|
|
74 |
|
|
# delete last subfield delimiter |
75 |
|
|
$sf = "" if ($sf =~ /\^\w*$/); |
76 |
|
|
} |
77 |
|
|
if (%v) { |
78 |
|
|
push @values, %v; |
79 |
|
|
} elsif ($sf && $sf !~ /^(\^\w)*\s*$/) { |
80 |
|
|
# regex above remove empty subfields |
81 |
|
|
push @values, $sf; |
82 |
|
|
} |
83 |
|
|
|
84 |
|
|
next if (! @values); |
85 |
|
|
|
86 |
|
|
if (! $num) { |
87 |
|
|
$num=$marc->createrecord(); |
88 |
|
|
$new++; |
89 |
|
|
} |
90 |
|
|
$marc->addfield({record=>$num, |
91 |
|
|
field=>$fld, |
92 |
|
|
i1=>" ", i2=>" ", |
93 |
|
|
value=>\@values}); |
94 |
|
|
|
95 |
|
|
} |
96 |
|
|
|
97 |
|
|
} |
98 |
|
|
} |
99 |
|
|
$total += $new; |
100 |
|
|
printf "\t%d (%0.2f%%) t: %d\n",$new,($new*100/$maxmfn),$total; |
101 |
|
|
} |
102 |
|
|
|
103 |
dpavlin |
229 |
$marc->output({file=>"> $marc_file",'format'=>"usmarc"}) |